Breast cancer is often curable if discovered in its early stages. Machine learning algorithms are used to automate disease identification. To increase the likelihood of recognizing the condition at an early stage, an effective classifier for automated diagnosis of breast cancer is needed. This research attempts to introduce a mixed machine learning model for predicting breast cancer. To outperform other strategies, it requires enormous amounts of data. Moreover, the intricate data models increase the training cost. For enhancing the efficiency of basic classifiers, ensemble learning holds great promise. A rapid discrete wavelet transforms using the wrapping approach is then used to extract the characteristics from the Region of Interest (ROI) mammography pictures. In addition to being too high to be categorized, the extracted wavelet coefficients also have incredibly high run-time complexity. The modified whale optimization method, which employs swarm intelligence, has been presented to lower time complexity and select the significant characteristics. Ensemble learning has been applied for the requisite efficiency in the suggested technique. Three machine learning (ML) classifiers—the extreme gradient boost classifier (XGB), support vector machine (SVM), and enhanced granular neural network (E-GNN)—form the ensemble voting system.